Saved in:
Bibliographic Details
Main Authors: Xu, Xingguo, Liu, Zhanyu, Zhou, Weixiang, Gao, Yuansheng, Cao, Junjie, Wang, Yuhao, Luo, Jixiang, Zhang, Dell
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.00695
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912934379126784
author Xu, Xingguo
Liu, Zhanyu
Zhou, Weixiang
Gao, Yuansheng
Cao, Junjie
Wang, Yuhao
Luo, Jixiang
Zhang, Dell
author_facet Xu, Xingguo
Liu, Zhanyu
Zhou, Weixiang
Gao, Yuansheng
Cao, Junjie
Wang, Yuhao
Luo, Jixiang
Zhang, Dell
contents Multi-modal object Re-Identification (ReID) aims to exploit complementary information from different modalities to retrieve specific objects. However, existing methods often rely on hard token filtering or simple fusion strategies, which can lead to the loss of discriminative cues and increased background interference. To address these challenges, we propose STMI, a novel multi-modal learning framework consisting of three key components: (1) Segmentation-Guided Feature Modulation (SFM) module leverages SAM-generated masks to enhance foreground representations and suppress background noise through learnable attention modulation; (2) Semantic Token Reallocation (STR) module employs learnable query tokens and an adaptive reallocation mechanism to extract compact and informative representations without discarding any tokens; (3) Cross-Modal Hypergraph Interaction (CHI) module constructs a unified hypergraph across modalities to capture high-order semantic relationships. Extensive experiments on public benchmarks (i.e., RGBNT201, RGBNT100, and MSVR310) demonstrate the effectiveness and robustness of our proposed STMI framework in multi-modal ReID scenarios.
format Preprint
id arxiv_https___arxiv_org_abs_2603_00695
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle STMI: Segmentation-Guided Token Modulation with Cross-Modal Hypergraph Interaction for Multi-Modal Object Re-Identification
Xu, Xingguo
Liu, Zhanyu
Zhou, Weixiang
Gao, Yuansheng
Cao, Junjie
Wang, Yuhao
Luo, Jixiang
Zhang, Dell
Computer Vision and Pattern Recognition
Multi-modal object Re-Identification (ReID) aims to exploit complementary information from different modalities to retrieve specific objects. However, existing methods often rely on hard token filtering or simple fusion strategies, which can lead to the loss of discriminative cues and increased background interference. To address these challenges, we propose STMI, a novel multi-modal learning framework consisting of three key components: (1) Segmentation-Guided Feature Modulation (SFM) module leverages SAM-generated masks to enhance foreground representations and suppress background noise through learnable attention modulation; (2) Semantic Token Reallocation (STR) module employs learnable query tokens and an adaptive reallocation mechanism to extract compact and informative representations without discarding any tokens; (3) Cross-Modal Hypergraph Interaction (CHI) module constructs a unified hypergraph across modalities to capture high-order semantic relationships. Extensive experiments on public benchmarks (i.e., RGBNT201, RGBNT100, and MSVR310) demonstrate the effectiveness and robustness of our proposed STMI framework in multi-modal ReID scenarios.
title STMI: Segmentation-Guided Token Modulation with Cross-Modal Hypergraph Interaction for Multi-Modal Object Re-Identification
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.00695